5G NETWORK TRAFFIC CLASSIFICATION SYSTEM USING INDIVIDUAL MACHINE LEARNING METHODS BASED ON NETWORK MONITORING TOOLS DATA

• Main Author: Shofiyyah Nur Pirusita, Nauroha
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/82258
• Institution: Institut Teknologi Bandung

In this era of fast evolving communication, 5G networks provide various advantages in the telecommunications business, including the promise of high- speed and high-capacity data connections. These advantages complicate network traffic management, necessitating accurate classification for efficient utilization across several applications and services. Manually approaches and protocol- based traffic identification system frequently fail to manage categorization efficiently, affecting network performance, service quality, and user experience. The goal of this project is to use individual machine learning algorithms to create an effective and efficient categorization system based on data from network monitoring tools. This method enables the development of more accurate and adaptive classification models to deal with changing traffic patterns. The developed individual machine learning algorithms are Support Vector Machine (SVM) and k-Nearest Neighbor (kNN). These techniques construct machine learning models by training using classification features on preprocessed data using a 5G traffic data approach derived from network monitoring tools. Subsequently, users can access visualizations and testing results based on the feature selections of dataset and machine learning model through a website. The testing results demonstrate that both machine learning models achieve high classification scoring (accuracy, precision, recall, f1-score, and mean cross validation), with each reaching approximately 99%. This classification method has the potential to be utilized as a solution for improving 5G network management and service quality.